(Pnpase) Reveals Differences in RNA Binding and Response to Phosphate Compared to Its Bacterial and Chloroplast Counterparts

Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Analysis of the human polynucleotide phosphorylase (PNPase) reveals differences in RNA binding and response to phosphate compared to its bacterial and chloroplast counterparts

VICTORIA PORTNOY, GILI PALNIZKY, SHLOMIT YEHUDAI-RESHEFF, FABIAN GLASER, and GADI SCHUSTER Department of Biology, Technion–Israel Institute of Technology, Haifa 32000, Israel

ABSTRACT PNPase is a major exoribonuclease that plays an important role in the degradation, processing, and polyadenylation of RNA in prokaryotes and organelles. This phosphorolytic processive enzyme uses inorganic phosphate and nucleotide diphosphate for degradation and polymerization activities, respectively. Its structure and activities are similar to the archaeal exosome complex. The human PNPase was recently localized to the intermembrane space (IMS) of the mitochondria, and is, therefore, most likely not directly involved in RNA metabolism, unlike in bacteria and other organelles. In this work, the degradation, polymerization, and RNA-binding properties of the human PNPase were analyzed and compared to its bacterial and organellar counterparts. Phosphorolytic activity was displayed at lower optimum concentrations of inorganic phosphate. Also, the RNA-binding properties to ribohomopolymers varied significantly from those of its bacterial and organellar enzymes. The purified enzyme did not preferentially bind RNA harboring a poly(A) tail at the 39 end, compared to a molecule lacking this tail. Several site-directed mutations at conserved amino acid positions either eliminated or modified degradation/polymerization activity in different manners than observed for the Escherichia coli PNPase and the archaeal and human exosomes. In light of these results, a possible function of the human PNPase in the mitochondrial IMS is discussed. Keywords: polyadenylation; RNA degradation; PNPase; human; phosphorolysis

INTRODUCTION 39 end processing, but recently a substantial degree of activity in the polymerization of heteropolymeric tails has also been Polynucleotide phosphorylase (PNPase) (EC 2.7.7.8) was reported (Mohanty and Kushner 2000; Mohanty et al. the first enzyme to be identified that catalyzes the forma- 2004). Moreover, PNPase is suggested to be the major tion of polynucleotides from ribonucleotides. However, polyadenylating enzyme in spinach chloroplasts, cyanobac- unlike RNA polymerases, PNPase neither requires a tem- teria, and Gram-positive bacteria (Yehudai-Resheff et al. plate nor can transcribe one. When a mixture of ribonu- 2001; Rott et al. 2003; Slomovic et al. 2006). PNPase was cleotide diphosphates (NDPs) serves as a substrate for the also reported to be a global regulator of virulence and polymerization reaction, the ensuing polymerization reac- persistence in Salmonella enterica (Clements et al. 2002). tion forms a random copolymer. The crystallographic structure analysis of PNPase from the PNPase catalyzes both processive 39 / 59 phosphorol- bacteria, Streptomyces antibioticus, revealed a homotri- ysis and 59 / 39 polymerization of RNA (Littauer and Soreq meric, circular-shaped complex (Symmons et al. 2000, 1982; Grunberg-Manago 1999; Littauer and Grunberg- 2002). Part of the PNPase population in E. coli is associated Manago 1999). In Escherichia coli, PNPase is mostly active with the endoribonuclease, RNase E, an RNA helicase, in 39 / 59 phosphorolysis during RNA degradation and enolase, and possibly other proteins in a high-molecular weight complex termed the degradosome (Marcaida et al. Reprint requests to: Gadi Schuster, Department of Biology, Technion– 2006). However, PNPase in the chloroplast was revealed to Israel Institute of Technology, Haifa 32000, Israel; e-mail: gadis@tx. be a homotrimeric complex without the association of technion.ac.il; fax: 972-4-8295587. Article published online ahead of print. Article and publication date are other proteins (Baginsky et al. 2001). It is also possible that at http://www.rnajournal.org/cgi/doi/10.1261/rna.698108. two homotrimeric complexes are associated together in

RNA (2008), 14:297–309. Published by Cold Spring Harbor Laboratory Press. Copyright Ó 2008 RNA Society. 297 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al. spinach chloroplasts, as PNPase elutes from a size-exclu- There is no obvious RNA substrate for the human sion column at about 600 kDa (Baginsky et al. 2001). PNPase in the IMS of the mitochondria, and its function The amino acid sequences of PNPases from bacteria, as and possible RNA target are still unknown. As the first step well as from the nuclear genomes of plants and mammals, to reveal the biological function of this enzyme in human display a high level of identity and feature similar structures cells, we considered biochemical characterization in com- composed of five motifs (Yehudai-Resheff et al. 2003; parison to the homologous bacterial and chloroplast Leszczyniecka et al. 2004). There are two core domains enzymes. To this end, the RNA phosphorolysis, binding, that display different degrees of homology with the E. coli and complex formation of the human PNPase were phosphorylase RNase PH, an a-helical domain between the analyzed. We found that when expressed in E. coli and two core domains, and two adjacent RNA-binding domains purified the human PNPase is active in both polymeriza- called KH and S1. These RNA-binding domains were tion and degradation and, like the bacterial enzyme, forms characterized in other RNA-binding proteins. The domain a trimeric homopolymer. However, the human PNPase structure of the bacterial and organellar PNPases is very differs from the bacterial and chloroplast enzymes in its well conserved in the exosomes of archaea and eukaryotic optimal inorganic phosphate concentration for degradation cells (Buttner et al. 2006; Houseley et al. 2006; Lin-Chao activity as well as its RNA-binding properties. Several site- et al. 2007). Indeed, the archaeal exosome is a phosphoro- directed mutations that eliminated the phosphorolytic lytic complex that is very similar to PNPase and, in systems activity of the E. coli enzyme or the archaeal and human in which it exists, is responsible for the degradation and exosomes have different effects on the human PNPase. polymerization of RNA, resembling the PNPase found in cyanobacteria (Portnoy et al. 2005; Portnoy and Schuster RESULTS 2006). The core structure of the eukaryotic exosome, which is responsible for the 39 / 59 exonucleolytic degradation of Human PNPase has a similar structure to that of the RNA in the nucleus and cytoplasm is also very similar to bacterial enzyme the PNPase and the archaeal exosome. While the human complex contains a phosphorolytic activity site, it could The human PNPase belongs to an evolutionarily conserved not be detected in the yeast exosome (Liu et al. 2006; enzyme family whose members were characterized in Dziembowski et al. 2007). The E. coli and chloroplast bacteria and organelles (Yehudai-Resheff et al. 2003; PNPases display high binding affinity to poly(A); the Leszczyniecka et al. 2004). This enzyme contains an N- responsible site is located in the S1 domain (Yehudai- terminal transit peptide of 37 amino acids that guides the Resheff et al. 2003). This phenomenon is important for its protein into the mitochondrial IMS and is, thereafter, function in the polyadenylation-stimulated degradation removed by cleavage (Chen et al. 2006; Rainey et al. pathway, in which the addition of poly(A) or poly(A)-rich 2006). Similar to its bacterial and organellar counterparts, sequences induces rapid exonucleolytic degradation (Lisitsky the protein is composed of two core domains homologous et al. 1997; Lisitsky and Schuster 1999). The enzymatic to the bacterial phosphorylase/exoribonuclease RNase PH activity is significantly enhanced by inorganic phosphate at (Fig. 1). The phosphorylase activity site was located at the concentrations of 10–20 mM, and is sensitive to secondary second core domain in the bacteria, Streptomyces antibio- structures in the RNA substrate (Yehudai-Resheff et al. ticus (Symmons et al. 2000) as well as in the homologous 2001, 2003). archaeal exosome protein Rrp41 (Lorentzen et al. 2005). In The human PNPase was identified in an overlapping spinach chloroplasts, in addition to the activity of the pathway screen to discover genes displaying coordinated second core domain, the first core domain is active in the expression as a consequence of terminal differentiation and degradation of polyadenylated RNA while the situation in senescence of melanoma cells (Leszczyniecka et al. 2002). E. coli is still unknown (Yehudai-Resheff et al. 2003). The Unlike chloroplast and plant mitochondria, in which first core domain is followed by an a-helical domain and PNPase is located in the stroma and matrix, the human the second core domain, by KH and S1 domains, which PNPase is mostly or exclusively located in the mitochon- were shown in the bacterial and chloroplast enzymes to drial intermembrane space (IMS) (Chen et al. 2006; Rainey be involved in RNA binding (Fig. 1; Jarrige et al. 2002; et al. 2006). In addition, PNPase was purified as an inter- Symmons et al. 2002; Yehudai-Resheff et al. 2003; Stickney acting protein with TCL1, a nonenzymatic oncoprotein et al. 2005). The S1 and KH domain motifs characterize promoting B- and T-cell malignancies and apoptotic many other RNA-binding proteins (Worrall and Luisi 2007). stimuli caused PNPase mobilization following cytochrome In order to observe the putative structure of the human c release to the cytoplasm (Chen et al. 2006; French et al. PNPase, we used the coordinates of the homologous 2007). Considering its IMS location, it most likely does not enzyme from the bacteria, S. antibioticus, that was deci- play a major or direct role in mitochondrial RNA metab- phered by crystallization and X-ray diffraction analysis olism, contrary to its role in bacteria, chloroplast, and plant (Symmons et al. 2000). Figure 1B presents the predicted mitochondria (Slomovic et al. 2006). trimeric structure compared to the bacterial enzyme.

298 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase

The human PNPase exhibits a very similar structure to the bacterial enzyme, having a doughnut-shaped trimer with a central channel that is of the correct dimensions to accommodate a single- stranded RNA molecule (Symmons et al. 2000). The circular structure is composed of the first and second core domains, while the KH and S1 RNA- binding domains are located at the top, adjacent to the gate entrance into the circle (Symmons et al. 2000; Lorentzen and Conti 2005; Lorentzen et al. 2005; Buttner et al. 2006). In Figure 1B, in which the trimer structure is presented as viewed from above, the KH domain (in yellow) is shown, located above the circular structure. To achieve a detailed analysis of the degradation, polymerization, and RNA- binding activities of the human PNPase, the protein was expressed in bacteria and purified to apparent homogeneity (Fig. 2A). In addition, several amino acids that are highly conserved when aligning the sequences of known PNPases (Fig. 1C), as well as the corresponding subunits of the archaeal, human, and yeast exosomes, which were predicted to be involved in the formation of the phosphorolysis site or the trimeric complex, based on the structural analysis, were modified by site-directed mutagenesis (Fig. 1B,C; Symmons et al. 2000; Lorentzen and Conti 2005; Lorentzen et al. 2005). The corresponding proteins were expressed in bacteria and purified also (Fig. 2A). Several additional conserved residues were modified as well; however, the corresponding proteins FIGURE 1. Human and bacterial PNPases and the mutated versions used in this work. (A) were not obtained in a soluble form The domain structures of the bacterial and human PNPases are presented schematically. The boxes represent the different domains as indicated, and the relative locations of the site- when expressed in bacteria. directed mutated amino acids are given. TP indicates the mitochondrial intermembrane transit peptide. (B) The resolved structure and predicted model of PNPase enzymes from Streptoomyces antibioticus (bacteria) and human (human) are shown. The trimeric dough- Phosphorolytic activity nut-like shape is presented in an orientation enabling easy observation of the central channel. of the human PNPase Each monomer is colored differently, and each domain is colored in one monomer of the enzyme, as shown in the scheme presented at the top. The homology-based modeling was As a phosphorylase, PNPase incorpo- z performed using the Phyre Server at http://www.sbg.bio.ic.ac.uk/ 3dpssm/. The complex was rates Pi and ADP in degradation and built by applying the crystal symmetry of the structure using the PyMOL program. A closeup polymerization processes, respectively view of the locations of the modified amino acids and their particular locations is presented below.(C) Alignment of the amino acid sequences of the conserved parts of human (Hs), E. coli (Littauer and Grunberg-Manago 1999). (Ec) and S. antibioticus (Sa) PNPases, as well as the hyperthermophilic archaea Sulfolobus When a 24-nucleotide (nt) RNA mole- solfataricus (Sso), human (hRrp45 and hRrp41), and yeast (yRrp45 and yRrp41) exosomes cule was incubated with the enzyme in proteins. The first and second signs indicate the first and second core RNase PH-like domains the presence of P and without the for PNPase, respectively. The corresponding protein names are indicated for the exosomes. i The yellow background shows the mostly conserved amino acids, which are colored red, and addition of ADP, only the degradation those mutated in this work are marked by red squares. of the RNA was observed (Fig. 2B, left

www.rnajournal.org 299 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al. panel). However, when the same substrate was incubated and nonpolyadenylated transcripts related to the ubiquitin with the enzyme in the presence of ADP but without Pi, mRNA were mixed together and assayed for degradation, both degradation and polymerization activities were the same rate was observed for both substrates, indicating detected (Fig. 2B, middle panel). When both ADP and Pi no competition of the polyadenylated RNA (Fig. 4E). In were present at the reaction mixture, the direction of contrast, competition between polyadenylated and non- activity, either polyadenylation or degradation, was depen- polyadenylated transcripts, resulting with the stabilization dent on their relative concentrations (Fig. 2B, left panel). of the nonpolyadenylated one, has been described pre- As expected of a phosphorylase, the byproduct of the viously for the E. coli and spinach chloroplast PNPases degradation of [32P]UTP-labeled RNA was identified as (Lisitsky et al. 1997; Lisitsky and Schuster 1999; Yehudai- UDP and the reaction was significantly stimulated by Pi Resheff et al. 2003). Therefore, we concluded that, unlike (Fig. 2C). Optimal degradation activity was obtained at the bacteria and chloroplast PNPases, the human enzyme does relatively low concentration of 0.1 mM (Fig. 2D). This not degrade polyadenylated RNA preferentially. This con- optimal concentration differs significantly from that of clusion was strengthened by the observation described in bacterial and chloroplast PNPases, in which the degrada- the continuation of the manuscript that human PNPase tion reaction was stimulated at much higher Pi concen- does not display high binding affinity to poly(A). Our trations of 10–20 mM (Littauer and Grunberg-Manago results differ from a previous analysis of this protein in 1999; Yehudai-Resheff et al. 2001; Jarrige et al. 2002). which preferential degradation of polyadenylated RNA was Indeed, at a Pi concentration of 10 mM, the degradation detected (French et al. 2007). activity of the human PNPase was severely inhibited (Fig. In order to determine the degradation efficiency that 2C; data not shown). The low activity observed here the human PNPase displays when incubated with RNA without the addition of Pi was eliminated when CaCl2 molecules containing extensive structural features, we was included in the reaction mixture in order to quench analyzed the RNAI of E. coli. This RNA, which oversees traces of Pi that were present in the reaction (cf Fig. 2B,C, the plasmid replication in the bacterial cell, is characterized which were performed without the addition of CaCl2, and Fig. 2D, in which 50 mM CaCl2 was added). This result suggested that the human PNPase can respond to much lower concentrations of Pi, changing its mode of activity between degradation and polymerization, compared to bacterial and chloroplast enzymes. The specificity of the enzyme for the polymerization reaction is, like that of the E. coli PNPase, high for ADP, with much less activity when incubated with other NDPs. No activity was observed with ATP nor the other NTPs, as well as mono phosphate nucleotides (Fig. 3).

Human PNPase degrades polyadenylated and nonpolyadenylated RNA at similar rates The PNPases of bacteria and chloroplasts preferentially degrade and polyadenylate RNA molecules containing a poly(A) tail at the 39 end. This preferential activity on polyadenylated transcripts has been attributed to the high binding affinity to poly(A) tracks located at the S1 domain in the spinach PNPase (Lisitsky et al. 1997; Lisitsky and Schuster 1999; Yehudai-Resheff et al. 2003). To analyze whether the human PNPase shares this preferential activity, RNA related to a region of the mitochondrial COX1 gene, either with or without the addition of a 49 nt poly(A) tail, was incubated with the human PNPase in either optimal polymerization or degradation conditions. The results revealed no differences in either the polymerization or the degradation activities (Fig. 4). Similar results were obtained with polyadenylated or nonpolyadenylated RNA substrates of varying lengths, derived from additional genes (Fig. 4E; data not shown). Moreover, when polyadenylated FIGURE 2. (Legend on next page)

300 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase by a tRNA-like structure, shown to hamper degradation by Cruz et al. 2002; Schubert et al. 2004; Stickney et al. 2005; bacterial PNPases, to a certain degree (Kaberdin et al. 1996). Amblar et al. 2007), an apparent sigmoid curve, typical of When the human or E. coli PNPases were incubated at the cooperative binding, was observed for the human enzyme. same concentration with the RNAI transcript, it was success- The Kd observed for the human PNPase was z16 nM. fully degraded, resulting in a very similar pattern of degra- Next, we examined whether the human PNPase, similar dation products (Fig. 5). We concluded that once titrated to to other PNPases analyzed to date, binds poly(A) with conditions for optimal degradation activity in terms of the higher affinity than poly(G) and poly(C). To this end, concentrations of Pi andADP,boththehumanandE. coli competition UV-cross-linking assays, in which increasing PNPases similarly degrade structured RNA molecules. amounts of the corresponding ribohomopolymer compete 32 with [ P]-(GU)12 RNA for binding of the protein, were carried out. The efficiency by which the ribohomopolymer Human PNPase displays different RNA-binding competes for protein binding reflects its affinity. Surpris- properties compared to bacterial ingly, while, as previously reported, the E. coli enzyme and organellar PNPases displayed high binding affinities to poly(U) and poly(A) Bacterial and spinach PNPase proteins were previously (Lisitsky and Schuster 1999), the human PNPase bound analyzed and displayed high binding affinity to poly(A) poly(U) and poly(G) with high affinity and displayed very and poly(U), in correlation with this enzyme’s necessity low binding affinity to poly(A) (Fig. 6B,C). The low affinity to compete for polyadenylated RNA (Lisitsky et al. 1997; to poly(A) sequences is in line with the observation that, Lisitsky and Schuster 1999). In order to determine the unlike the case of the E. coli and chloroplast PNPases, the binding curve, RNA-binding of the purified protein was human enzyme displayed no preferential activity for poly- analyzed in a UV-cross-linking assay using increasing adenylated RNA (Fig. 4). amounts of PNPase. As shown in Figure 6A, while the E. coli enzyme displayed a saturation binding curve with a K d Unlike the spinach chloroplast PNPase, the human of 11 nM, in agreement with previous reports (Bermudez- PNPase does not bind the E. coli enzyme to form a heterotrimeric complex when expressed in E. coli When expressing the recombinant spinach PNPase protein FIGURE 2. Polymerization and degradation activities of human in E. coli, we previously found that a heterotrimeric PNPase. (A) Coomassie-stained polyacrylamide gel profile of the functional complex, consisting of both spinach and E. coli recombinant proteins following expression in E. coli.Proteinswere subunits, was formed (Yehudai-Resheff et al. 2003). In this purified by nitrilotriacetic acid (NTA) agarose affinity chromatography, anion-exchange (MonoQ), and heparin affinity chromatography steps, case, the presence of bacterial subunits in the trimer inter- and loaded onto a 10% SDS-PAGE gel. Molecular mass markers (in fered with the determination of the activities and RNA- kDa) are shown at the left and right (M). Ec, E. coli PNPase; Hs, human binding properties of the recombinant spinach PNPase. We PNPase. In other lanes, the purified proteins of the corresponding then, therefore, expressed the spinach protein in an E. coli mutants as labeled by the modified amino acid residue are shown. (B) 32 Human PNPase was incubated with [ P]-RNA (GU)12 oligonucleotide strain in which the endogenous PNPase gene was inter- in the presence of either Pi (0.1 mM) or ADP (2.5 mM), as indicated in rupted by the genomic insertion of DNA and therefore, the the figure, in order to promote degradation or polyadenylation activity, bacterial protein was not expressed (Lopez et al. 1999). In respectively. Samples were removed at 0, 5, 15, 30, and 60 min, and the RNA was analyzed by denaturing PAGE and autoradiography. The the present work, since we were concerned that a similar input RNA, as well as polymerized and partially degraded products, are phenomenon could hamper characterization of the human indicated. In the right panel, incubation was for 30 min in the present of PNPase, we initially expressed the protein in the E. coli 2.5 mM ADP and Pi at the concentration indicated on the top.NP-no strain in which PNPase is not expressed. In order to protein. (C) Thin-layer chromatography (TLC) analysis of the degradation products. Human PNPase was incubated with uniformly [32P] examine whether, similar to the spinach PNPase, the UTP-labeled RNA corresponding to part of the human mitochondrial human enzyme can form a heterotrimeric complex com- transcript COX1. Following the incubation, the reaction products were posed of E. coli and human PNPase subunits, the human spotted onto a polyethyleneimine TLC plate, which was developed with PNPase was produced in nonmutated E. coli strain express- 0.9 M GnCl, dried, and autoradiographed. The PNPase activity was ing the intrinsic protein. The recombinant his6 tagged assayed either in the absence (lane 1) or presence of 0.1 mM Pi (lane 2). In lane 3, UMP and UDP were analyzed as markers on the same TLC human PNPase was purified by affinity chromatography plate and visualized by fluorescence quenching. The dotted lines mark and analyzed on SDS-PAGE in order to see if an associated the boundaries of the fluorescence pattern. In lane 4, a control reaction E. coli PNPase coeluted with it. As shown in Figure 7, the in which the RNA was incubated without the addition of protein, was result revealed that no bacterial PNPase copurified with analyzed. (D) Human PNPase activity in response to Pi concentrations. 32 [ P]-labeled RNA oligonucleotide (GU)12 was incubated for 15 min the his6 tagged human PNPase, indicating that a heterotri- with human PNPase at different Pi concentrations, as indicated. CaCl2 meric PNPase complex was not obtained (Fig. 7A, lane 3). (50 mM) was added to the reaction mixture in order to precipitate the However, when spinach chloroplast PNPase was expressed Pi already present in the sample. Following incubation, the RNA was analyzed by denaturing PAGE and autoradiography. No enzyme was in this strain, the E. coli polypeptide copurified with the added to the reaction in the lanes labeled (–). recombinant spinach protein (Fig. 7A, lane 4). In order to

www.rnajournal.org 301 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al.

mutagenesis (Fig. 1). The corresponding modified proteins were then analyzed for activity and RNA-binding (Fig. 8). Although other conserved amino acids, such as G509DandG513R, were modified as well, the corresponding proteins could not be obtained when expressed in bacteria in a soluble form and could there- FIGURE 3. Of the NDPs, human PNPase displays a preference for ADP during polymeri- fore not be analyzed. 32 zation activity. [ P]-labeled RNA oligonucleotide (GU)12 was incubated with human (A)or Changing the aspartic acid residue at E. coli (B) proteins and 2.5 mM of each of the corresponding nucleotides, as indicated at the position 544, which is involved in the top of the figure. For the E. coli enzyme (B), only the NDPs nucleotides were assayed. Samples formation of the catalytic site, to glycine were withdrawn at 10 min, and the RNA analyzed by denaturing PAGE and autoradiography. The input RNA and polymerized products are shown on the left. The input RNA (lane [–]) as (D544G), significantly reduced degrada- well as a control reaction in which no protein was added (lane NP) were also analyzed. tion activity but surprisingly enhanced polymerization. In contrast, in the E. coli PNPase, the equivalent change, (D492G), verify that even a minute amount of E. coli PNPase did not resulted in the loss of phosphorolytic activity but also bind the human polypeptide, we analyzed the purified abolished polyadenylation activities by 90% (Table 1; Jarrige recombinant proteins with specific antibodies. No signal et al. 2002). Similar results of enhancing polymerization and was obtained with antibodies specific to the E. coli PNPase significantly inhibiting degradation activities were obtained when probing the purified human PNPase (Fig. 7B, left when arginines, 445 and 446, involved in RNA binding at panel, lane 2). We concluded that, unlike the spinach the catalytic site, were converted to glutamic acid chloroplast PNPase, human polypeptides cannot form a (R445,446E). This observation is interesting, since modifying heterotrimer with the E. coli protein. the corresponding conserved arginines in the E. coli PNPase, (R398,399D), the archaeal exosome, (R98,99E of Rrp41), and human exosome, (R94V,K95A of Rrp41), resulted in the Recombinant human PNPase forms a homotrimeric complete inhibition of phosphorolytic activity for E. coli complex; formation is impeded by a D135G mutation PNPase and the human exosome, as well as the degradation located at the first core activity for the archaeal complex (Table 1; Jarrige et al. 2002; Lorentzen et al. 2005; Liu et al. 2006). Changing serine Next, we analyzed whether human PNPase, like the E. coli 484 and aspartic acid 538 to alanine (S GandD A) enzyme, forms a homotrimer of z240 kDa (Amblar et al. 484 538 practically eliminated both the polyadenylation and degra- 2007). Alternatively, the spinach chloroplast enzyme is dation activities of the human PNPase (Fig. 8). When the characterized as a 580-kDa complex that is possibly corresponding residues in the human and archaeal exo- comprised of two homotrimers (Symmons et al. 2000; somes, involved in the formation of the P binding and the Baginsky et al. 2001). Analysis of the complex formation of i catalytic active sites, were modified, the result was elimina- the various enzymes by fractionation on a size-exclusion tion of activity, as well (Table 1). column revealed that the human enzyme migrated at a Interestingly, changing aspartic acid 135, which is located molecular weight of 230 kDa, similar to E. coli and unlike in the first core domain, to glycine, (D G), a mutation the spinach chloroplast homolog (Fig. 7C). The same size 135 shown above to cause the inability to form the homotrimer was measured previously, upon fractionation on a blue- complex (Fig. 7), resulted in the almost complete inhibition native gel (Chen et al. 2006). A site-directed mutant, in of degradation and polyadenylation activities. The corre- which one amino acid located in the first core domain was sponding mutation in the E. coli PNPase, (D G), resulted changed (D G), was characterized by the inability to form 96 135 with almost complete loss of the degradation activity but the homotrimer complex, and the protein was eluted from only slightly reduced the polymerization activity (Table 1; the column in a monomeric form of z90 kDa (Fig. 7C). Jarrige et al. 2002). The RNA-binding property of the modified proteins, as disclosed by the UV-cross-linking assay, was drastically Changing conserved amino acids resulted in the reduced compared to the wild-type and the E. coli proteins altering of phosphorolytic and RNA-binding activities even in the D544GandR445,446E mutants, in which the In order to determine which amino acids are important for polymerization activity was significantly enhanced (Fig. 8C). the phosphorolytic and RNA-binding activities of PNPase, Together, the mutational analyses revealed both similarities several residues that are evolutionarily conserved between and differences between the activities of the mutated bacterial and organellar PNPases, as well as the archaeal, PNPase, compared with parallel mutations in the E. coli yeast, and human exosomes, were modified via site-directed enzyme and the archaeal and human exosomes (Table 1).

302 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase

acids sequence of the S1 domain of the E. coli, spinach chloroplast, and human PNPase revealed significant amounts of conserved amino acids but not drastic differences that could possibly account for the diverse binding properties (Fig. 9A). Nevertheless, when the S1 domain of the human enzyme was replaced with that of the E. coli PNPase, high binding affinity to poly(A) was observed, indicating that the S1 domain is responsible for this phenomenon (Fig. 9B). Interestingly, the high binding affinity to poly(G), a characteristic of human but not E. coli PNPase, was still observed in the human PNPase harboring the E. coli S1 domain (Fig. 9B). This result suggests that the high affinity to poly(G) characteristic of the human PNPase is not located at the S1 domain.

DISCUSSION In this study, the phosphorolytic activity and RNA-binding properties of the human PNPase were studied and compared to bacterial and organellar PNPases. In addition, conserved amino acids in PNPases and the archaea and human exosomes, which were shown to be located at the activity sites, were mutated, and the modified proteins analyzed for activity and RNA-binding properties. The human PNPase was found to be active in the phosphorolysis of RNA, resembling its bacterial and organellar

FIGURE 4. Similar activities of human PNPase on nonpolyadenylated and polyadenylated substrates. [32P] RNA corresponding to part of the mitochondrial gene COX1, either without (A,C) or with the addition of 49 adenosins at the 39 end (B and D) was incubated with the protein and 0.1 mM Pi (C,D) or 2.5 mM ADP (A,B). In panel E, degradation assay was performed to non-polyadenylated ubiquitin transcript (I), the same transcript with the addition of 64 adenosines (II), and the two transcripts mixed together (III). Samples were withdrawn at 0, 15, 30, and 120 min, and the RNA analyzed by denaturing PAGE and autoradiography.

Domain swapping of the S1 domain between the E. coli and human PNPases, conferred high poly(A) binding affinity to the human enzyme Previously, we have found that the S1 domain of the spinach chloroplast PNPase is responsible for high binding affinity to poly(A) (Yehudai-Resheff et al. 2003). This phenomenon is important for the activity of this enzyme in bacteria, chloroplast, and plant mitochondria in the FIGURE 5. Similar degradation activities of human and E. coli PNPases. [32P] RNA corresponding to the E. coli RNAI (111 nt) was polyadenylation stimulated RNA degradation pathway incubated with the human or E. coli PNPases (100 ng each). Pi was (Lisitsky et al. 1997; Lisitsky and Schuster 1999; Yehudai- added to a final concentration of 0.1 or 10 mM for the human and Resheff et al. 2003). The human PNPase is located in the E. coli enzymes, respectively. Samples were collected at 0, 5, 15, 30, 60, mitochondria IMS and does not seems to be directly 90, and 120 min, and the RNA was analyzed by denaturing PAGE and autoradiography. RNA markers were fractionated in the lanes marked involved in this pathway, and accordingly, displayed low M1 and M2. The length in nucleotides is shown on the right. The binding affinity to poly(A) (Fig. 6). Alignment of the amino predicted secondary structure of RNAI is shown to the right.

www.rnajournal.org 303 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al.

homologs, as previously described (Hayakawa and Sekigu- chi 2006; Sarkar and Fisher 2006; French et al. 2007). Differences in the Pi concentrations for optimal activity and RNA-binding properties, between the human PNPase and other PNPases, were detected as well. The characterization of the human PNPase is of great interest because of its distinct location in the human cell in comparison to plant organellar PNPases. While in plant organelles PNPase is located in the mitochondrial matrix and the chloroplast stroma, the human PNPase is located in the IMS of the mitochondrion, where no RNA has yet been observed. This unique location is probably reflected by diverse functions acquired in the evolution of plant and animal PNPases. In bacteria, PNPase is directly involved in the degradation, processing, and nontemplate polymerization of RNA (Kushner 2004; Deutscher 2006). Indeed, in cyanobacteria, PNPase is exclusively responsible for the polyadenylation activity of the RNA degradation pathway, and deletion of the PNPase gene is lethal (Rott et al. 2003). Similarly, in hyperthermophilic and methanogenic archaea, the archaeal exosome, which is very similar to PNPase, is responsible for both polyadenylation and degradation of RNA (Portnoy et al. 2005; Portnoy and Schuster 2006). PNPase also plays a direct role in the degradation and processing of organellar transcripts in the mitochondria and chloroplast of plants (Walter et al. 2002; Perrin et al. 2004a,b). However, no PNPase exists in yeast; possibly a general case for all fungi (Dziembowski et al. 2003). Interestingly, it was recently observed that the yeast exosome exhibits no phosphorolytic activity, implying that with neither PNPase nor a phosphorolytic exosome, no RNA phosphorolytic activity occurs in yeast (Liu et al. 2006; Dziembowski et al. 2007). As described above, the human PNPase was recently found to be located in the IMS of mitochondria and is, therefore, excluded from the matrix where the mitochondrial transcripts are present (Chen et al. 2006; Rainey et al. 2006). This observation suggests that it is not directly FIGURE 6. RNA-binding properties of human and E. coli PNPases. (A) Human PNPase displays a sigmoidal curve which suggests involved in the metabolism of mitochondrial RNA, con- cooperative binding. Human and E. coli PNPases (10 ng each) were trary to the case of plants. On the other hand, silencing the analyzed for RNA-binding by a UV light cross-linking assay using expression of PNPase in human cells was found to increasing amounts of PNPase, as indicated in the figure inset. significantly affect, perhaps indirectly, the amount and The radioactive signals obtained in three independent experiments (error bars are represented) were quantified and plotted in order length of stable poly(A) tails which characterized the 39 to determine the shape of the binding curve and the Kd observed. ends of mitochondrial transcripts (Slomovic and Schuster The black line with circled pointsrepresentsthehumanPNPase 2008). Therefore, and since there is not an obvious known and the dashed line with filled points represents the E. coli RNA substrate in the IMS, it was of great interest to analyze PNPase. The Kd observed for E.coli and human PNPases were 11 and 16 nM, respectively. (B) High affinity binding of human the phosphorolytic and RNA-binding properties of the PNPase to poly(U)andpoly(G). Human or E. coli PNPases were human PNPase in comparison to the bacterial and plant 32 incubated with [ P]-RNA corresponding to (GU)12, and a homo- PNPases that are directly involved in RNA metabolism. ribopolymer at a molar excess, as shown in the figure. Following UV cross-linking and ribonuclease digestion, the label transfer from RNA to protein was analyzed by SDS-PAGE and autoradiography. Differences between human and previously InthelanelabeledB, no protein was added to the assay as a control. (C) Quantization of the competition assays. The UV light cross- characterized PNPases linking competition assays were performed as described for (B), and The RNA degradation activity of human PNPase has been the intensity of the UV light cross-linking band without competitor was defined as 100%. Data shown are the average of at least three previously shown for purified recombinant or in vitro experiments. translated protein (Leszczyniecka et al. 2002; Sarkar et al.

304 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase

organellar PNPases is required for activity because of their direct involvement in the polyadenylation-stimulated RNA degradation pathway. Due to its binding preference to polyadenylated RNA, following the polyadenylation of RNA by poly(A)-polymerase or PNPase, PNPase binds and degrades the polyadenylated molecule. Thus, polyadenylated RNA molecules compete with nonpolyadenylated molecules for PNPase binding (or additional exoribonucleases) in bacteria and plant organelles (Schuster et al. 1999; Regnier and Arraiano 2000; Dreyfus and Regnier 2002; Slomovic et al. 2006). Since, in human mitochondria PNPase is mostly or exclusively not located in the matrix, it is most likely not directly involved in a polyadenylation- stimulated RNA degradation pathway, and therefore high binding affinity to poly(A) is not required. All characterized PNPases bind the ribohomopolymer, poly(U), with high affinity. However, while bacterial and chloroplast PNPases exhibited low binding affinity to poly(G) (Lisitsky and Schuster 1999; Yehudai-Resheff et al. 2003), the human enzyme binds this ribohomopolymer with high affinity.

FIGURE 7. Unlike the spinach chloroplast PNPase, the human protein does not form a complex with its E. coli counterpart. (A) Human (lanes 1 and 3), E. coli (lane 2), or spinach chloroplast (lane 4) PNPases were expressed in E. coli strain in which the expression of the endogenous PNPase was knocked-out (lane 1), or in E. coli strain containing the endogenous PNPase (lanes 2, 3, and 4). The expressed proteins were purified as described in Materials and Methods and analyzed on SDS-PAGE followed by silver staining. The identities of the different PNPases are indicated on the right.TheE. coli endogenous PNPase was copurified with the spinach protein (lane 4) but not with the human one (lane 3). (B) Recombinant human PNPase was expressed and purified in E. coli strains either lacking (lane 1)or containing (lane 2) endogenous PNPase. In lane 3, a purified recombinant E. coli PNPase was loaded. The proteins were analyzed by SDS-PAGE and immunoblotted to a nitrocellulose membrane that was incubated with antibodies against the E. coli PNPase or the human PNPase, as indicated in the figure. (C) Recombinant human PNPase forms a high molecular weight complex of z300 kDa. Recombinant proteins (z5 mg) were fractionated on a Superdex 200 size-exclusion column. The elution profile of molecular-weight markers: Thyroglobulin 665 kDa, Feritin 440 kDa, Catalase 232 kDa, Aldolase 158 kDa, and Albumin 67 kDa, fractionated on the same column, is indicated at the top. FIGURE 8. Polymerization, degradation, and RNA-binding activities of the different mutants. (A,B)[32P]-labeled RNA oligonucleotide (GU)12 was incubated with wild-type (wt) PNPase or site-directed mutants as indicated. Reactions contained 2.5 mM of ADP and 2005; French et al. 2007). The results of the present work incubated for 10 min in (A) and 0.1 mM Pi incubating for 15 min in disclosed that the human PNPase exhibits phosphorolytic (B). RNA was purified and analyzed by denaturing PAGE and activity that could efficiently polymerize or degrade all of autoradiography. The input RNA, as well as the polymerized and the RNA molecules that were examined. There are two degradation products, are indicated on the left. A control reaction in which no protein was added is shown in lane NP. (C) Wild-type (wt) major differences between the human and the thus far human PNPase and its site-directed mutants (10 ng each) were characterized bacterial and organellar PNPases. The first analyzed for RNA-binding by the UV light cross-linking assay using 32 difference is that while the bacterial and plant organellar [ P] RNA corresponding to the (GU)12. Following UV cross-linking enzymes disclosed a high binding affinity to poly(A), the and ribonuclease digestion, the label transfer from RNA to protein was analyzed by SDS-PAGE and autoradiography. In the lane labeled human PNPase binds poly(A) with very low affinity (Fig. NP, no protein was added to the assay as a control. Ec, purified E. coli 6). The high affinity poly(A)-binding of bacterial and plant PNPase.

www.rnajournal.org 305 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al.

Another possibility is that, similar to TABLE 1. Changes in polymerization and degradation activities of site-directed mutants compared with the corresponding mutations in E. coli PNPase (Jarrige et al. 2002), the S. endonuclease G, PNPase does not have solfataricus (archaea) exosome (Buttner et al. 2005; Lorentzen and Conti 2005; Lorentzen a designated function within the IMS; et al. 2005), and human exosome (Liu et al. 2006) core proteins rather, it is stored in this compartment, released during apoptosis, and functions in the phosphorolysis of a substrate found in the cytoplasm (Chen et al. 2006). Further studies concerning the biological and molecular properties of the human PNPase will explore these hypotheses and reveal the biological function of this enzyme.

MATERIALS AND METHODS

Production of recombinant PNPase and its mutated versions The corresponding DNA sequence of the The corresponding residues of the yeast exosome that is phosphorolyticly inactive are shown as well (Liu et al. 2006; Dziembowski et al. 2007). A minus sign (–) indicates no mature protein (without the transit peptide) activity detected. A plus sign (+) indicates that the corresponding activity was detected. The was amplified by PCR using the primers numbers of plus signs indicate the relative strength of the activity. ND, activity not GGTACCCGAGCTGTGGCCGTGGACTTA determined; Pol., polymerization activity; Deg., RNA degradation activity. and GTCGACTCACTGAGAATTAGATGA TGACTGTGA and the PNPase cDNA, kindly obtained from P.B. Fisher (Columbia University). The Acc65I and SalI restriction Whether or not this is related to its biological function is sites that were added are bold and underlined. The different point yet to be revealed. mutations in the full-length (wild-type [wt]) protein were The second difference is the optimal concentration of Pi introduced using the site-directed mutagenesis kit (Stratagene). for degradation activity. While the E. coli and chloroplast To create the chimera protein in which the S1 domain of human PNPase was replaced by that from the E. coli enzyme, a BamHI site PNPases display optimal activity at Pi concentrations exceeding 10 mM, the human PNPase optimal activity was introduced between KH and S1 domains using a site-directed was found at a concentration of 0.1 mM and is actually mutagenesis kit (Stratagene). The E. coli S1 domain, to which the same restriction site was introduced, was cloned using this site and inhibited at higher concentrations. While the P concen- i the XhoI restriction site located in the pet20b vector. tration is believed to be 10–20 mM in bacteria and the For expression in E. coli, the PCR products were inserted into organellar matrixes, its concentration in the mitochondrial the Pet 20b vector with the addition of a His6 tag at the C IMS in human cells is unknown. It could be suggested that terminus. The proteins were expressed in the BL-21 DE3 pLysS either the human enzyme acquired optimal activity at a low or the strain ENS134-3 lacking the endogenous PNPase and Pi concentration for its function within the IMS, or that containing the T7 RNA polymerase [BL21(DE3)(lacZ:Tn10 other PNPases were modified during evolution to function malPp534:PT7lacZ-Arg5)(pnp:Tn5)] (Lopez et al. 1999), kindly at higher Pi concentrations in bacteria and within the obtained from Marc Dreyfus (Ecole Normale Supe´rieure). The organellar matrix. proteins were purified according to the manufacturer’s protocol There are two possibilities to explain the IMS location of using a NTA-agarose affinity column with additional purification this enzyme: There may be an RNA substrate in the IMS steps using MonoQ and Heparin columns (Pharmacia). The wt that is subjected to the phosphorolytic activity of PNPase and mutant D91G proteins were purified additionally by fraction- which plays an important function in the fine tuning of the ation on the size-exclusion column Superdex 200. All of the NDP/P concentrations inside the mitochondria, which proteins were purified to the stage of one Coomassie-stained band i on SDS-PAGE (Fig. 1), and no contaminating ribonuclease activity may also influence the NTP concentrations. In human cells was detected. In order to examine possible RNA content of the in which PNPase was silenced by RNAi, significant effects PNPase preparation, RNA was extracted from a purified fraction upon the length and amount of poly(A) tails, which containing 50 mg of PNPase and analyzed by gel fractionation and characterize mitochondrial transcripts, were detected, silver staining. Less than 1 ng of RNA was detected (data not together with changes in ATP concentration (Slomovic shown). In addition, incubation of the enzyme with Pi for 30 min and Schuster 2008). Furthermore, proper ATP concentra- (phosphorolysis) or extensive RNase A treatment did not reveal tions are essential for efficient polyadenylation and tran- any differences. Therefore, the purified PNPase contains only scription of mitochondrial RNA (Dubin et al. 1982). minor traces, if any, of RNA. Specific antibodies for the E. coli and

306 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase

omitted. Following incubation, the RNA was isolated and analyzed by denaturing PAGE and autoradiography. For a thin-layer chromatography analysis of the degradation products, an aliquot of each sample was spotted on a polyethyleneimine thin layer chromatography plate, which was then developed with 0.9 M GnCl, dried, and exposed to autoradiography (Yehudai- Resheff et al. 2003). Nucleosides di-, and three phosphates (5 mg of each) were sepa- rated on the same plate and visualized by fluorescence quenching.

UV light cross-linking assay UV light cross-linking of proteins to radiolabeled RNA was performed as previously FIGURE 9. Changing the S1 domain of human PNPase to that of the E. coli enzyme resulted described (Lisitsky et al. 1997). The proteins with high binding affinity to poly(A). (A) Alignment of the amino acid sequences of the S1 32 domain of E. coli (Ec_S1), spinach chloroplast (So_S1), and human (Hs_S1) PNPases. were mixed with [ P]-RNA in a buffer Conserved amino acids are indicated with a gray background. (B) High-affinity binding of the containing 10 mM HEPES-NaOH, pH 7.9, chimera protein to poly(U), poly(A), and poly(G). The chimera protein, as schematically 30 mM KCl, 6 mM MgCl2, 0.05 mM EDTA, 32 presented, was incubated with [ P]-RNA corresponding to (GU)12, and a homoribopolymer 2 mM DTT, and 8% glycerol and cross- at a molar excess, as shown in the figure. Following UV cross-linking and ribonuclease linked immediately at a UV light cross-linker digestion, the label transfer from RNA to protein was analyzed by SDS-PAGE and that was set at 1.8 J (Hoefer). This was autoradiography. In the lane labeled B, no protein was added to the assay as a control. (C) Quantization of the competition assays. The UV light cross-linking competition assays were followed by digestion of the RNA with 10 mg performed as described for (B), and the intensity of the UV light cross-linking band without of RNase A and 30 units of RNase T1 at competitor was defined as 100%. 37°C for 1 h. The proteins were then fractionated by SDS-PAGE and analyzed by autoradiography. For the competition assay, human PNPases were kindly obtained from Agamemnon the protein was first mixed with the ribohomopolymers and then J. Carpousis (Centre National de la Recherche Scientifique) and the radiolabeled RNA was added. Paul B. Fisher (Columbia University), respectively. Structure prediction and complex model building Synthetic RNAs and antibodies Homology-based modeling of the three-dimensional structure of cDNAs encoding the 100 nt of the 39 end of the human human PNPase was performed using the Phyre Server at http:// mitochondrial transcript COX1 with or without the addition of www.sbg.bio.ic.ac.uk/z3dpssm/. The visualization of the human 49 adenosines were described (Slomovic et al. 2005). Part of the PNPase trimer complex was performed by inserting the mono- cDNA encoding 267 nt of the ubb (ubiquitin), with or without the meric model described above into a pseudorhombohedral (H32) addition of 62 adenosines was used to generate the corresponding space group with dimensions similar to those of the Streptomyces transcript. These cDNAs were cloned to the pGEM vector, and the antibioticus crystal structure (PDB code 1E3H) using PyMOL RNAs were transcribed using T7 RNA polymerase and radioac- (Delano). tively labeled with [32P]UTP (Portnoy and Schuster 2006). The full-length transcription products were purified from 5% denaturing polyacrylamide gels. The plasmid for the transcription of ACKNOWLEDGMENTS the E. coli RNAI was described previously (Portnoy and Schuster We thank Paul B. Fisher and Piotr Stepien for the human PNPase 2006). The (GU) oligonucleotide was radioactively labeled at the 12 cDNA and antibodies and Marc Dreyfus for the PNPase minus 59 end using [32P] g-ATP and polynucleotide kinase. strain of E. coli. We thank Agamemnon J. Carpousis for E. coli PNPase antibodies. We also thank Shimyn Slomovic for critical Polyadenylation and degradation assays reading and editing the manuscript and Rafal Tomecki for help with the RNA-binding experiments. This work was supported by Polyadenylation and degradation activities of the recombinant grants from the Israel Science Foundation (ISF), the United proteins were assayed as previously described (Yehudai-Resheff States–Israel Binational Science Foundation (BSF), and the United et al. 2003). [32P]-RNA was incubated with the corresponding States–Israel Binational Agricultural Research and Development proteins in buffer E (20 mM HEPES, pH 7.9, 60 mM KCl, 12.5 Fund (BARD). S.Y.-R. was supported by the Ministry of Absorp- mM MgCl2, 0.1 mM EDTA, 2 mM DTT, and 17% glycerol) at tion fellowship. 37°C with the addition of Pi. When polyadenylation was assayed, ADP or the corresponding nucleotide was added and the Pi Received June 21, 2007; accepted October 24, 2007.

www.rnajournal.org 307 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Portnoy et al.

REFERENCES Leszczyniecka, M., Kang, D.C., Sarkar, D., Su, Z.Z., Holmes, M., Valerie, K., and Fisher, P.B. 2002. Identification and cloning of Amblar, M., Barbas, A., Gomez-Puertas, P., and Arraiano, C.M. 2007. human polynucleotide phosphorylase, hPNPase old-35, in the The role of the S1 domain in exoribonucleolytic activity: Substrate context of terminal differentiation and cellular senescence. Proc. specificity and multimerization. RNA 13: 317–327. Natl. Acad. Sci. 99: 16636–16641. Baginsky, S., Shteiman-Kotler, A., Liveanu, V., Yehudai-Resheff, S., Leszczyniecka, M., DeSalle, R., Kang, D.C., and Fisher, P.B. 2004. The Bellaoui, M., Settlage, R.E., Shabanowitz, J., Hunt, D.F., origin of polynucleotide phosphorylase domains. Mol. Phylogenet. Schuster, G., and Gruissem, W. 2001. Chloroplast PNPase exists Evol. 31: 123–130. as a homo-multimer enzyme complex that is distinct from the Lin-Chao, S., Chiou, N.T., and Schuster, G. 2007. The PNPase, Escherichia coli degradosome. RNA 7: 1464–1475. exosome, and RNA helicases as the building components of Bermudez-Cruz, R.M., Garcia-Mena, J., and Montanez, C. 2002. evolutionarily conserved RNA degradation machines. J. Biomed. Polynucleotide phosphorylase binds to ssRNA with same affinity Sci. 14: 523–532. as to ssDNA. Biochimie 84: 321–328. Lisitsky, I. and Schuster, G. 1999. Preferential degradation of poly- Buttner, K., Wenig, K., and Hopfner, K.P. 2005. Structural framework adenylated and polyuridinylated RNAs by bacterial exoribonu- for the mechanism of archaeal exosomes in RNA processing. Mol. clease polynucleotide phosphorylase (PNPase). Eur. J. Biochem. Cell 20: 461–471. 261: 468–474. Buttner, K., Wenig, K., and Hopfner, K.P. 2006. The exosome: A Lisitsky, I., Kotler, A., and Schuster, G. 1997. The mechanism of pre- macromolecular cage for controlled RNA degradation. Mol. ferential degradation of polyadenylated RNA in the chloroplast: Microbiol. 61: 1372–1379. The exoribonuclease 100RNP/PNPase displays high binding affin- Chen, H.W., Rainey, R.N., Balatoni, C.E., Dawson, D.W., Troke, J.J., ity for poly(A) sequence. J. Biol. Chem. 272: 17648–17653. Wasiak, S., Hong, J.S., McBride, H.M., Koehler, C.M., Littauer, U.Z. and Grunberg-Manago, M. 1999. Polynucleotide Teitell, M.A., et al. 2006. Mammalian polynucleotide phosphory- phosphorylase. In: The encyclopedia of molecular biology (ed. T.E. lase is an intermembrane space RNase that maintains mitochon- Creighton), pp. 1911–1918. Wiley, New York. drial homeostasis. Mol. Cell. Biol. 26: 8475–8487. Littauer, U.Z. and Soreq, H. 1982. Polynucleotide phosphorylase. Clements, M.O., Eriksson, S., Thompson, A., Lucchini, S., Academic Press, New York. Hinton, J.C., Normark, S., and Rhen, M. 2002. Polynucleotide Liu, Q., Greimann, J.C., and Lima, C.D. 2006. Reconstitution, phosphorylase is a global regulator of virulence and persistency in activities, and structure of the eukaryotic RNA exosome. Cell Salmonella enterica. Proc. Natl. Acad. Sci. 99: 8784–8789. 127: 1223–1237. Deutscher, M.P. 2006. Degradation of RNA in bacteria: Comparison Lopez, P.J., Marchand, I., Joyce, S.A., and Dreyfus, M. 1999. The C- of mRNA and stable RNA. Nucleic Acids Res. 34: 659–666. doi: terminal half of RNase E, which organizes the Escherichia coli 10.1093/nar/gkj472. degradosome, participates in mRNA degradation but not rRNA Dreyfus, M. and Regnier, P. 2002. The poly(A) tail of mRNAs: processing in vivo. Mol. Microbiol. 33: 188–199. Bodyguard in eukaryotes, scavenger in bacteria. Cell 111: Lorentzen, E. and Conti, E. 2005. Structural basis of 39 end RNA 611–613. recognition and exoribonucleolytic cleavage by an exosome RNase Dubin, D.T., Montoya, J., Timko, K.D., and Attardi, G. 1982. PH core. Mol. Cell 20: 473–481. Sequence analysis and precise mapping of the 39 ends of HeLa Lorentzen, E., Walter, P., Fribourg, S., Evguenieva-Hackenberg, E., cell mitochondrial ribosomal RNAs. J. Mol. Biol. 157: 1–19. Klug, G., and Conti, E. 2005. The archaeal exosome core is a Dziembowski, A., Piwowarski, J., Hoser, R., Minczuk, M., hexameric ring structure with three catalytic subunits. Nat. Struct. Dmochowska, A., Siep, M., van der Spek, H., Grivell, L., and Mol. Biol. 12: 575–581. Stepien, P.P. 2003. The yeast mitochondrial degradosome. Its Marcaida, M.J., DePristo, M.A., Chandran, V., Carpousis, A.J., and composition, interplay between RNA helicase and RNase activities Luisi, B.F. 2006. The RNA degradosome: Life in the fast lane of and the role in mitochondrial RNA metabolism. J. Biol. Chem. adaptive molecular evolution. Trends Biochem. Sci. 31: 359–365. 278: 1603–1611. Mohanty, B.K. and Kushner, S.R. 2000. Polynucleotide phosphorylase Dziembowski, A., Lorentzen, E., Conti, E., and Seraphin, B. 2007. A functions both as a 39 to 59 exonuclease and a poly(A) polymerase single subunit, Dis3, is essentially responsible for yeast exosome in Escherichia coli. Proc. Natl. Acad. Sci. 97: 11966–11971. core activity. Nat. Struct. Mol. Biol. 14: 15–22. Mohanty, B.K., Maples, V.F., and Kushner, S.R. 2004. The Sm-like French, S.W., Dawson, D.W., Chen, H.W., Rainey, R.N., Sievers, S.A., protein Hfq regulates polyadenylation dependent mRNA decay in Balatoni, C.E., Wong, L., Troke, J.J., Nguyen, M.T., Koehler, C.M., Escherichia coli. Mol. Microbiol. 54: 905–920. et al. 2007. The TCL1 oncoprotein binds the RNase PH domains Perrin, R., Lange, H., Grienenberger, J.M., and Gagliardi, D. 2004a. of the PNPase exoribonuclease without affecting its RNA degrad- AtmtPNPase is required for multiple aspects of the 18S rRNA ing activity. Cancer Lett. 248: 198–210. metabolism in Arabidopsis thaliana mitochondria. Nucleic Acids Grunberg-Manago, M. 1999. Messenger RNA stability and its role in Res. 32: 5174–5182. doi: 10.1093/nar/gkh852. control of gene expression in bacteria and phages. Annu. Rev. Perrin, R., Meyer, E.H., Zaepfel, M., Kim, Y.J., Mache, R., Genet. 33: 193–227. Grienenberger, J.M., Gualberto, J.M., and Gagliardi, D. 2004b. Hayakawa, H. and Sekiguchi, M. 2006. Human polynucleotide Two exoribonucleases act sequentially to process mature 39-ends phosphorylase protein in response to oxidative stress. Biochemistry of atp9 mRNAs in Arabidopsis mitochondria. J. Biol. Chem. 279: 45: 6749–6755. 25440–25446. Houseley, J., LaCava, J., and Tollervey, D. 2006. RNA-quality control Portnoy, V. and Schuster, G. 2006. RNA polyadenylation and by the exosome. Nat. Rev. Mol. Cell Biol. 7: 529–539. degradation in different Archaea; Roles of the exosome and RNase Jarrige, A., Brechemier-Baey, D., Mathy, N., Duche, O., and R. Nucleic Acids Res. 34: 5923–5931. doi: 10.1093/nar/gkl763. Portier, C. 2002. Mutational analysis of polynucleotide phosphor- Portnoy, V., Evguenieva-Hackenberg, E., Klein, F., Walter, P., ylase from Escherichia coli. J. Mol. Biol. 321: 397–409. Lorentzen, E., Klug, G., and Schuster, G. 2005. RNA polyadeny- Kaberdin, V.R., Chao, Y.H., and Lin-Chao, S. 1996. RNase E cleaves at lation in Archaea: Not observed in Haloferax while the exosome multiple sites in bubble regions of RNA I stem–loops yielding polyadenylates RNA in Sulfolobus. EMBO Rep. 6: 1188–1193. products that dissociate differentially from the enzyme. J. Biol. Rainey, R.N., Glavin, J.D., Chen, H.W., French, S.W., Teitell, M.A., Chem. 271: 13103–13109. and Koehler, C.M. 2006. A new function in translocation for the Kushner, S.R. 2004. mRNA decay in prokaryotes and eukaryotes: mitochondrial i-AAA protease Yme1: Import of polynucleotide Different approaches to a similar problem. IUBMB Life 56: phosphorylase into the intermembrane space. Mol. Cell. Biol. 26: 585–594. 8488–8497.

308 RNA, Vol. 14, No. 2 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Activities of the human PNPase

Regnier, P. and Arraiano, C.M. 2000. Degradation of mRNA Slomovic, S., Portnoy, V., Liveanu, V., and Schuster, G. 2006. RNA in bacteria: Emergence of ubiquitous features. Bioessays 22: polyadenylation in prokaryotes and organelles; Different tails tell 235–244. different tales. Crit. Rev. Plant Sci. 25: 65–77. Rott, R., Zipor, G., Portnoy, V., Liveanu, V., and Schuster, G. 2003. Stickney, L.M., Hankins, J.S., Miao, X., and Mackie, G.A. 2005. RNA polyadenylation and degradation in cyanobacteria are similar Function of the conserved S1 and KH domains in polynucleotide to the chloroplast but different from Escherichia coli. J. Biol. Chem. phosphorylase. J. Bacteriol. 187: 7214–7221. 278: 15771–15777. Symmons, M.F., Jones, G.H., and Luisi, B.F. 2000. A duplicated fold Sarkar, D. and Fisher, P.B. 2006. Polynucleotide phosphorylase: An is the structural basis for polynucleotide phosphorylase catalytic evolutionary conserved gene with an expanding repertoire of activity, processivity, and regulation. Structure 8: 1215–1226. functions. Pharmacol. Ther. 112: 243–263. Symmons, M.F., Williams, M.G., Luisi, B.F., Jones, G.H., and Sarkar, D., Park, E.S., Emdad, L., Randolph, A., Valerie, K., and Carpousis, A.J. 2002. Running rings around RNA: A superfamily Fisher, P.B. 2005. Defining the domains of human polynucleotide of phosphate-dependent RNases. Trends Biochem. Sci. 27: 11–18. phosphorylase (hPNPaseOLD-35) mediating cellular senescence. Walter, M., Kilian, J., and Kudla, J. 2002. PNPase activity determines Mol. Cell. Biol. 25: 7333–7343. the efficiency of mRNA 39-end processing, the degradation of Schubert, M., Edge, R.E., Lario, P., Cook, M.A., Strynadka, N.C., tRNA and the extent of polyadenylation in chloroplasts. EMBO J. Mackie, G.A., and McIntosh, L.P. 2004. Structural characterization 21: 6905–6914. of the RNase E S1 domain and identification of its oligonucleo- Worrall, J.A. and Luisi, B.F. 2007. Information available at cut rates: tide-binding and dimerization interfaces. J. Mol. Biol. 341: 37–54. Structure and mechanism of ribonucleases. Curr. Opin. Struct. Schuster, G., Lisitsky, I., and Klaff, P. 1999. Update on chloroplast Biol. 17: 128–137. molecular biology: Polyadenylation and degradation of mRNA in Yehudai-Resheff, S., Hirsh, M., and Schuster, G. 2001. Polynucleotide the chloroplast. Plant Physiol. 120: 937–944. phosphorylase functions as both an exonuclease and a poly(A) Slomovic, S. and Schuster, G. 2008. Stable PNPase RNAi silencing: Its polymerase in spinach chloroplasts. Mol. Cell. Biol. 21: 5408–5416. effect on the processing and adenylation of human mitochondrial Yehudai-Resheff, S., Portnoy, V., Yogev, S., Adir, N., and Schuster, G. RNA. RNA (this issue). doi: 10.1261/rna.697308. 2003. Domain analysis of the chloroplast polynucleotide phos- Slomovic, S., Laufer, D., Geiger, D., and Schuster, G. 2005. Poly- phorylase reveals discrete functions in RNA degradation, polyadenylation and degradation of human mitochondrial RNA: The adenylation, and sequence homology with exosome proteins. Plant prokaryotic past leaves its mark. Mol. Cell. Biol. 25: 6427–6435. Cell 15: 2003–2019.

www.rnajournal.org 309 Downloaded from rnajournal.cshlp.org on October 6, 2021 - Published by Cold Spring Harbor Laboratory Press

Analysis of the human polynucleotide phosphorylase (PNPase) reveals differences in RNA binding and response to phosphate compared to its bacterial and chloroplast counterparts

Victoria Portnoy, Gili Palnizky, Shlomit Yehudai-Resheff, et al.

RNA 2008 14: 297-309

References This article cites 52 articles, 20 of which can be accessed free at: http://rnajournal.cshlp.org/content/14/2/297.full.html#ref-list-1

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the Service top right corner of the article or click here.

To subscribe to RNA go to: http://rnajournal.cshlp.org/subscriptions